Although test anxiety is typically negatively related to performance on cognitive ability tests, little research has systematically investigated whether differences in test anxiety result in measurement bias on cognitive ability tests. The current paper uses a structural equation modeling technique to explicitly test for measurement bias due to test anxiety. In comparing two nested models, one hypothesizing measurement bias and one not, results show the model indicating no measurement bias to be statistically more acceptable. Consistent with the deficits model of test anxiety, results show that test anxiety is associated primarily with only the g-factor and has no significant influence on observed test scores. (Contains 2 figures and 4 tables.)